Flowering Plant Families: An Overview

James L. Reveal

ONE of the purposes of the Flora of North America project is to present a series of established names at all ranks for extant vascular plants north of Mexico, and thus by its very nature, this work will be a standard by which all future modifications will be judged. The problems are more acute at the generic and specific ranks, but major changes will undoubtedly occur also at the ranks of family and order during the useful life of this and the following volumes.

This is not an indictment of modern systems of classification, especially that proposed by Cronquist and adopted here (A.Cronquist 1981, 1988). Rather, the state of our knowledge regarding the circumscriptions of vascular plant taxa at higher ranks will continue to undergo revision, as a result of improvements suggested by an ever-expanding group of biologists who are using new techniques to address questions of relationship.

The longstanding difficulty that these researchers will encounter, and one that the present work is certain to promote, is the problem of acceptance of any change in the established norm. The wide acceptance of the de Candolle and Lindley systems during the first half of the nineteenth century certainly slowed the adoption of changes suggested by Bentham and Hooker in the 1860s. In contrast, the changes proposed by Engler and Prantl in the last two decades of the 1800s were readily adopted because, by then, the need for change was recognized. Moreover, the detailed studies by Engler and Prantl and their associates justified acceptance of the proposed modifications. The Engler and Prantl model remained the norm until the late 1950s when Cronquist and Takhtajan provided a series of innovative changes that were accepted because, once again, a need for substantial change was recognized, and sound rationales were presented for that change. During each era, including the present, other views were presented (Airy Shaw in J.C. Willis 1973; R.M.T. Dahlgren 1983; R.M.T. Dahlgren et al. 1985; G.Dahlgren 1989, 1989b; A.Goldberg 1986, 1989; V.H. Heywood 1978; J.Hutchinson 1959, 1973; D.J. Mabberley 1987; H.Melchior 1964; E.Rouleau 1981; G.L. Stebbins 1974; A.L. Takhtajan 1987; R.F. Thorne 1983, 1992, 1992b), but at some point, one system began to dominate.

With this in mind, the following commentary is offered. It is a reminder to the reader that there are differing views as to circumscriptions and relationships. Finally, whereas classifications at the lower ranks can be evaluated readily in the herbarium or in the field, relationships at the family and ordinal ranks are more difficult to demonstrate simply because of the time required to gain sufficient knowledge.

Cronquist (1981, 1988) used 3 major superfamilial ranks within Magnoliophyta, the flowering plants (appendix 15.1): class, subclass, and order. In his system, the dicotyledons were placed in Magnoliopsida, the monocotyledons in Liliopsida. Within Magnoliopsida, he distinguished 6 subclasses and 64 orders; in Liliopsida, 5 subclasses and 19 orders. A total of 389 families were accepted, 323 of them dicots and 66 monocots. A.L. Takhtajan (1987) used 4 major superfamilial ranks: class, subclass, superorder, and order. Eight subclasses of dicotyledons were distinguished and 4 among the monocots. Within the dicots, Takhtajan accepted 36 superorders, 128 orders, and 429 families; within the monocots, he had 16 superorders, 36 orders, and 104 families. Recently Takhtajan (pers. comm. 1993) has acepted 450 dicot families arranged among 145 orders, 45 superorders, and 10 subclasses. R.F. Thorne (1992) currently recognizes 437 families (351 dicots) distributed among 28 superorders (19 dicots) and 71 orders (52 dicots). (See appendix 15.2 for a concordance of the families recognized by these three authors.)

The first subclass in the Cronquist system is Magnoliidae. Takhtajan divided this into two subclasses, Magnoliidae and Ranunculidae, and the first three superorders of the Thorne system basically encompass the same series of families. These three authors arranged the orders and families in differing linear sequences, although the overall relationships among the families are remarkably similar, with only a few exceptions (e.g., Takhtajan and Thorne placed Coriariaceae and Sabiaceae with the rosoids).

As to circumscriptions of the North American families included in Cronquist's Magnoliidae, some differences of opinion exist. Cronquist defined Ranunculaceae to include Glaucidiaceae and Hydrastidaceae, which both Takhtajan and Thorne recognized as separate. Cronquist and Takhtajan maintain Fumariaceae, which Thorne put in Papaveraceae.

Hamamelididae (spelled by Cronquist as "Hamamelidae") were treated by both Takhtajan and Thorne in two parts. Takhtajan defined Hamamelididae to include such North American families as Hamamelidaceae (plus Altingiaceae, which Cronquist retained in Hamamelidaceae), Platanaceae, Fagaceae, Betulaceae, Myricaceae, Juglandaceae, and Casuarinaceae. Takhtajan placed Ulmaceae, Moraceae, Cannabaceae, and Urticaceae in Dilleniidae. Takhtajan and Thorne included among their hamamelids two families that Cronquist placed in Rosidae: Buxaceae and Simmondsiaceae.

The placement of Leitneriaceae is uncertain. Cronquist included them within their own order in Hamamelididae, between his Urticales and Juglandales. Takhtajan placed the order in Rosidae next to Rutales, and Thorne included the family within Rutales. Most of the hamamelid families Cronquist recognized are circumscribed similarly by other recent workers.

Caryophyllidae, with the exception of Polygonaceae and Plumbaginaceae, are a tightly knit group, uniformly circumscribed by most recent workers. There is some disagreement over the circumscriptions of the various families.

The placement of Plumbaginaceae has been either near Caryophyllales or Primulales. Cronquist and Takhtajan accepted the former position, Thorne the latter. The isolated nature of Polygonaceae has not been questioned, and their association with Caryophyllales is based more on tradition than on firm evidence.

Cronquist's Dilleniidae, with 77 families, are the second largest subclass of his Magnoliopsida, and one whose representatives in our flora tend to be highly specialized. The families of the subclass are more narrowly defined by Takhtajan (who accepted 153), and they were somewhat more scattered among the rosoid families by Thorne. The overall arrangement of the families, however, is fairly similar.

The traditional inclusion of Paeoniaceae in Ranunculaceae is no longer accepted, with Cronquist referring the family to Dilleniidae. Takhtajan and Thorne placed Paeoniaceae near Glaucidiaceae (which Cronquist included within Ranunculaceae) in Ranunculidae.

Cronquist placed Symplocaceae in Ebenales, whereas Takhtajan and Thorne retained the family in Theales. Cronquist's placement of Sarraceniaceae, Nephenthaceae, and Droseraceae in a single order, Nepenthales, in Dilleniidae, has been questioned by monographers and others. Nonetheless, there is no firm opinion as to an acceptable alternative. Takhtajan retained Sarraceniaceae in Dilleniidae, Nepenthaceae in Magnoliidae, and Droseraceae in Rosidae. Thorne's placements have been equally diverse.

The Flacourtiaceae of Cronquist and most previous workers are the "garbage pail" of Dilleniidae. They were separated by Takhtajan into Berberidopsidaceae, Aphloiaceae, Kiggelariaceae, and Plagiopteraceae, but only the Plagiopteraceae were removed from the immediate vicinity of Flacourtiaceae.

Cronquist included the Loasaceae, without conviction, in Violales of Dilleniidae. Thorne and Takhtajan placed the family in its own superorder near Solanaceae in Asteridae. All three concurred that the Salicaceae belong to Dilleniidae (or their equivalent) and are not among the amentiferous families as defined by Engler and Prantl. The closeness of Tamaricaceae, Frankeniaceae, and Fouquieriaceae to Salicaceae was also accepted.

A consensus on the definition of Ericaceae has been gradually emerging, with the various families recognized through most of the 1800s now reduced primarily to Ericaceae, with the occasional recognition (as by Cronquist) of Pyrolaceae and Monotropaceae.

Rosidae, as defined by Cronquist, comprise 18 orders and 116 families, many being common in temperate regions although most have tropical distributions. Rosales are the basal element; they are a coherent group with little disagreement about their overall makeup. Opinions vary regarding the circumscriptions of families that Engler and Prantl related to Saxifragaceae, and regarding the relationship of Cornales (Cornaceae in a broad sense) to the woody members of Saxifragaceae. The circumscription of Grossulariaceae by Cronquist included most of the woody saxifrage genera. Other recent authors have restricted, more properly in my opinion, Grossulariaceae to the genus Ribes and have distinguished several additional families for North American plants, notably Escalloniaceae, Phyllonomataceae, Iteaceae, and Pterostemonaceae. The apparent similarity of woody saxifrage families with Cornales was noted by both Takhtajan and Thorne.

The circumscription of herbaceous Saxifragaceae is equally unsettled. Cronquist included Lepuropetalaceae, Parnassiaceae, and Penthoraceae in his Saxifragaceae, but others have preferred to recognize these as distinct.

The arguments for and against treating Rosaceae and Fabaceae (Fabales) as several families have been discussed since the early 1800s. Cronquist, Takhtajan, and Thorne all recognize Rosaceae in a broad sense. Cronquist has separated Mimosaceae and Caesalpiniaceae from Fabaceae, but both Takhtajan and Thorne included them in the Fabaceae.

Cronquist placed the Euphorbiaceae in Rosidae, along with Buxaceae and Simmondsiaceae. Takhtajan, however, placed Euphorbiaceae in a terminal position in his Dilleniidae, with Buxaceae and Simmondsiaceae placed in their own orders in Hamamelididae. Thorne retained Simmondsiaceae in Euphorbiales near Urticales, and moved Buxaceae into the equivalent of Hamamelididae.

Cronquist's placement of Rhamnales next to Eurphorbiales was accepted by Thorne but rejected by Takhtajan. Both Takhtajan and Thorne placed Elaeagnaceae in an order, Elaeagnales, situated between Santalales and Proteales. Cronquist placed Elaeagnaceae and Proteaceae in Proteales.

The circumscription of Zygophyllaceae was reviewed by Takhtajan, who recognized five families where Cronquist had but one; the additional North American name among the five is Peganaceae. Likewise, Takhtajan recognized five families in what Cronquist defined as Geraniaceae, but all of the segregates are beyond our flora.

The distinction between Araliaceae and Apiaceae has long been regarded as weak, with tradition being as significant as morphological characters in distinguishing the two families. Combining the two was promoted by Thorne (1983), who now suggests (1992, 1992b), as have many Old World workers, that if a distinction is made between Araliaceae and Apiaceae, then Hydrocotylaceae must be recognized as well. Takhtajan (pers. comm. 1993) now concurs.

The majority of the sympetalous dicotyledons were placed in Asteridae by Cronquist, who defined the group as a subclass of 11 orders and 49 families. Takhtajan, on the other hand, recognized two unequal groups, the bulk of the families going to Lamiidae, and Asteridae being restricted mainly to Campanulales and Asterales. In addition, Takhtajan referred Dipsacales to a terminal position within Rosidae.

Cronquist has held to traditional circumscriptions of the 49 families of the Asteridae. The families tend to have large numbers of species, with members found in a variety of both temperate and tropical habitats. Even when the families are more narrowly defined, as done by Takhtajan, they still tend to be large and widely distributed.

The basal group within Asteridae is Gentianales. Takhtajan divided Loganiaceae into three families, including Spigeliaceae, and placed Rubiaceae next to them, a position with which Thorne agreed. The traditional separation of Apocynaceae and Asclepiadaceae was maintained by Cronquist but not accepted by Thorne.

Takhtajan divided Cronquist's Solanaceae into three families, and he recognized two New World tropical families, Sclerophylacaceae and Goetzeaceae. Both Cronquist and Thorne separated Cuscutaceae from Convolvulaceae, but Thorne did not.

The diverse Scrophulariaceae and their allies were variously treated. Thorne and Takhtajan have retained Orobanchaceae within the Scrophulariaceae. Cronquist placed the Paulowniaceae in Bignoniaceae, but Takhtajan and Thorne put them in Scrophulariaceae. Selaginaceae, known only from cultivation in our range, was included in Globulariaceae by Cronquist, in Scrophulariaceae by Takhtajan, and was recognized as distinct by Thorne. Martyniaceae, placed in Pedaliaceae by Cronquist, were accepted as distinct by Takhtajan and Thorne.

Lobeliaceae were distinguished from Campanulaceae by Takhtajan, but treated within the Campanulaceae by both Cronquist and Thorne. In addition, Takhtajan adopted Nemacladaceae for three genera from western North America; he also distinguished Cyphiaceae and Cyphocarpaceae, two families beyond our flora.

Within Asteridae, the Caprifoliaceae and relatives are particularly complex. Thorne has moved Viburnaceae and Sambucaceae into an expanded Adoxaceae.

Notable studies have been done recently on the families in the Liliopsida, with the late Rolf Dahlgren in the forefront (R.M.T. Dahlgren et al. 1985). His premature death slowed progress toward understanding the evolution of the flowering plants.

Dahlgren et al. rejected the concept that the monocots had an aquatic ancestry, and their view was adopted by Thorne but not by Cronquist or Takhtajan. While this view does not alter the circumscription of any family, it does influence what is to be regarded as the basal taxon. Cronquist and Takhtajan considered this to be Alismatidae; Dahlgren and Thorne began with the equivalent of Liliidae.

The tradition of distinguishing monocots and dicots at one of the higher taxonomic ranks, as by Cronquist and Takhtajan at the rank of class, was not retained by Dahlgren and Thorne, who considered the flowering plants to be divided into a series of superorders. True, members of each of their superorders have either monocotyledonous or dicotyledonous features, and no one superorder includes both monocotyledonous and dicotyledonous members. Dahlgren and Thorne emphasized that there is no fundamental difference in these features and that monocotyledonous plants arose from several different dicotyledonous groups.

In the monocots, Cronquist maintained families with broad, traditional definitions, while Takhtajan generally circumscribed families more narrowly. The interrelationships posited for particular groups, ranking aside, are rather similar in the two schemes. Even the groupings proposed by Dahlgren and Thorne generally comprised the same complements of families. But there remains great disagreement regarding the lilioid families.

Cronquist defined Liliales as a taxon of 15 families and nearly 8000 species, with Liliaceae defined to include what Dahlgren considered to represent 4 orders (at least in part) and 27 families! In 1988 Cronquist labeled his Liliaceae "inchoate," and stated that he would divide them into several families if he could "find a reasonable way to do it." Takhtajan did attempt to find a reasonable way, and he distributed elements found in Cronquist's Liliaceae among 5 orders (Liliales, Amaryllidales, Asparagales, Dioscoreales, and Alstroemeriales) containing 23 families. Thorne now has recognized 3 orders (Liliales, Asparagales, and Dioscoreales) with 20 families.

The problems noted with Liliaceae are mirrored in the circumscriptions of other families such as Agavaceae, Aloaceae, and Smilacaceae. For the most part, resolution involves families found outside our area, but one may be noted here. Aloaceae were defined by Cronquist to exclude Asphodelaceae, which he placed in Liliaceae. Dahlgren, Thorne, and Takhtajan considered Aloaceae to be no more than a subfamily or tribe of Asphodelaceae.

Conclusion

Systems of classification attempt to show relationships and to provide a convenient means of expressing them. Such arrangements are difficult to substantiate, but consensus on at least some of them is possible.

The differences of opinion expressed by Cronquist, Takhtajan, and Thorne in their respective classifications are not as great as they might seem. Rarely does one author circumscribe a family to include taxa that another assigns to widely separated places. More frequently, a group of families will be associated by one author, but scattered about by another. In most of these instances, the families involved are highly specialized (e.g., Droseraceae, Nepenthales). Where there are truly complex differences of opinion (e.g., Grossulariaeae and Cornales), there is often no evident way to resolve them. While Cronquist has accepted only 388 families as compared to 533 recognized by Takhtajan, the difference is not so much a question of relationship as one of rank (J.L. Reveal 1993). The vast majority of segregate families accepted by Takhtajan are directly associated in his scheme with the families into which Cronquist has put them.

Revised and more sophisticated taxonomic schemes will certainly enhance our knowledge of plants. One of the strengths of the Linnaean system of botanical nomenclature, however, is its fundamental stability. Thus, despite altered family conceptions and revised family sequences, the names by which plants are called (i.e., genus and species names) nearly always remain the same. For example, it makes no difference to the name Desmanthus illinoensis, the Illinois bundleflower, if the plant is treated as a member of the Mimosaceae, or within a semidistinctive subfamily of the Fabaceae, the bean family. Similarly, the nomenclature within the genus Euphorbia (the spurges) is not affected whether its family, the Euphorbiaceae, is treated among the Rosidae, as in Cronquist, or among the Dilleniidae, as in Takhtajan.

Acknowledgments

This summary would not have been possible without the assistance of the late Arthur Cronquist, Armen L. Takhtajan, and Robert F. Thorne, who have kindly provided me with their most up-to-date information. In particular, Drs. Takhtajan and Thorne have provided me with information that they have yet to publish, and for permission to use their data I am most grateful. My colleague in the review of family nomemclature, Ruurd D. Hoogland, remains, as always, most helpful. Work on the nomenclature of vascular plant families is supported in part by National Science Foundation Grant BSR-8812816.